Re: [df-scan.c] Optimise DF_REFs ordering in collection_rec, use HARD_REG_SETs instead of bitmaps

[Dimitrios Apostolou <jimis at gmx dot net>]

On Fri, 8 Jul 2011, Paolo Bonzini wrote:
> On 07/08/2011 05:51 AM, Dimitrios Apostolou wrote:
> > +	  /* first write DF_REF_BASE */
>
> This is not necessary.  These uses are written to use_vec, while the uses 
> from REG_EQUIV and REG_EQUAL are written to eq_use_vec (see 
> df_ref_create_structure).

Thanks for pointing this out, I missed it, it's complex to follow 
what is written where. Perhaps there is meaning in changing the interface 
of df_ref_create_structure() to accept the particular vector.

> Also, anyway this wouldn't work because you would have to split the loop in 
> two.  I'll attribute that to the time of day when you were writing the 
> message. :)

Even though I'm having my own disputes with Morpheus, sleep deprivation 
wasn't tha culprit for this. :-) My intention was to keep DF_REF_BASEs as 
close to each other, because a close-to-be-sorted array will be sorted 
faster with the sorting function I'll write... :-)

But anyway as you've explained it's irrelevant since they are going 
to different vectors.


Thanks,
Dimitris